linux sdio card休眠处理 sdio card removed解决办法

最新推荐文章于 2022-11-26 15:07:07 发布

原创最新推荐文章于 2022-11-26 15:07:07 发布 · 置顶 · 6.8k 阅读

8 ·

CC 4.0 BY-SA版权

文章标签：

#sdio suspend 处理 #sdio card removed #dump_stack #linux #博通

驱动总结同时被 2 个专栏收录

49 篇文章

订阅专栏

APQ8064

16 篇文章

订阅专栏

本文详细介绍了在SDIO卡使用过程中遇到的card removed问题及解决方案，通过设置nonremovable和keeppower避免睡眠模式下的异常行为。

摘要生成于 C知道，由 DeepSeek-R1 满血版支持，前往体验 >

最近调试几款sdio card suspend时都会出现sdio card removed，之后

要么死机要么模块不能正常工作，根本原因也就是休眠没有处理好。昨天终于找到了

解决方法。

         1：host端需要设置nonremovable,软件设置：mmc->caps |=
MMC_CAP_NONREMOVABLE;
         2：host端需要设置keep power,在sdio card suspend时软件设
置:host->pm_flags | = MMC_PM_KEEP_POWER；

         目前我所接触到的4g网卡,博通网卡,sdio加密t卡等都会出现card
removed问题。

         关于2补充下需要在sdio card pm中置host keep power:

         static int XXX_suspend(struct device *pdev)
{
     struct sdio_func *func = dev_to_sdio_func(pdev);

     /* keep power while host suspended */
     ret = sdio_set_host_pm_flags(func, MMC_PM_KEEP_POWER);
     if (ret) {
         sd_err(("%s: error while trying to keep power\n", __FUNCTION__));
         return ret;
     }

     return 0;
}

static const struct dev_pm_ops XXX_pm_ops = {
     .suspend    = XXX_suspend,
     .resume        = XXX_resume,
};

今天花了点时间理清了card removed的原因。

出现1错误的原因是在resume 时pm_notify调用到了msm_rescan,如果没
有置nonremovable,将会跑进detect,而进去就会detect出了card removed,
进而进行了移除操作。

  void mmc_rescan(struct work_struct *work)
     {
         struct mmc_host *host =
             container_of(work, struct mmc_host, detect.work);
         bool extend_wakelock = false;

         if (host->rescan_disable)
             return;

         mmc_bus_get(host);

         /*
          * if there is a _removable_ card registered, check whether it is
          * still present
          */
         if (host->bus_ops && host->bus_ops->detect && !host->bus_dead
             && !(host->caps & MMC_CAP_NONREMOVABLE))
             host->bus_ops->detect(host);
         ---------
     }
         下面是dump_stack()打印的调用信息：
<4>[   48.709320] [<c0013d4c>] (unwind_backtrace+0x0/0x11c) from 
[<bf011d74>] (dhdsdio_disconnect+0xc/0x10c [bcmdhd])
<4>[   48.720214] [<bf011d74>] (dhdsdio_disconnect+0xc/0x10c [bcmdhd]) 
from [<bf01f9a8>] (bcmsdh_remove+0x1c/0x98 [bcmdhd])
<4>[   48.730041] [<bf01f9a8>] (bcmsdh_remove+0x1c/0x98 [bcmdhd]) from 
[<bf001a00>] (bcmsdh_sdmmc_remove+0x28/0x70 [bcmdhd])
<4>[   48.740600] [<bf001a00>] (bcmsdh_sdmmc_remove+0x28/0x70 [bcmdhd]) 
from [<c05855e0>] (sdio_bus_remove+0x38/0xf0)
<4>[   48.753021] [<c05855e0>] (sdio_bus_remove+0x38/0xf0) from 
[<c03a26c8>] (__device_release_driver+0x9c/0xe0)
<4>[   48.762725] [<c03a26c8>] (__device_release_driver+0x9c/0xe0) from 
[<c03a2728>] (device_release_driver+0x1c/0x28)
<4>[   48.771881] [<c03a2728>] (device_release_driver+0x1c/0x28) from 
[<c03a21f8>] (bus_remove_device+0x124/0x140)
<4>[   48.781829] [<c03a21f8>] (bus_remove_device+0x124/0x140) from 
[<c039fe30>] (device_del+0x108/0x16c)
<4>[   48.790802] [<c039fe30>] (device_del+0x108/0x16c) from 
[<c058596c>] (sdio_remove_func+0x1c/0x28)
<4>[   48.799560] [<c058596c>] (sdio_remove_func+0x1c/0x28) from 
[<c0584bfc>] (mmc_sdio_remove+0x3c/0x68)
<4>[   48.808593] [<c0584bfc>] (mmc_sdio_remove+0x3c/0x68) from 
[<c0584cb4>] (mmc_sdio_detect+0x8c/0xb4)
<4>[   48.817565] [<c0584cb4>] (mmc_sdio_detect+0x8c/0xb4) from 
[<c057e0c8>] (mmc_rescan+0x7c/0x2d8)
<4>[   48.826141] [<c057e0c8>] (mmc_rescan+0x7c/0x2d8) from [<c0099dd8>] 
(process_one_work+0x27c/0x484)
<4>[   48.834991] [<c0099dd8>] (process_one_work+0x27c/0x484) from 
[<c009a1f0>] (worker_thread+0x210/0x3b0)
<4>[   48.844177] [<c009a1f0>] (worker_thread+0x210/0x3b0) from 
[<c009e08c>] (kthread+0x80/0x8c)
<4>[   48.852355] [<c009e08c>] (kthread+0x80/0x8c) from [<c000eaa8>] 
(kernel_thread_exit+0x0/0x8)
<6>[   49.083587] mmc2: card 0001 removed

出现2错误的原因是在sdio resume时，如果没有设置keep power就会
跑sdio reset与go idle（CMD 0）后进入init_card第一个CMD就不会响应了，也就是
这2个操作后card异常了。

static int mmc_sdio_resume(struct mmc_host *host)
{
     int i, err = 0;

     BUG_ON(!host);
     BUG_ON(!host->card);

     /* Basic card reinitialization. */
     mmc_claim_host(host);

     /* No need to reinitialize powered-resumed nonremovable cards */
     if (mmc_card_is_removable(host) || !mmc_card_keep_power(host)) {
         sdio_reset(host);
         mmc_go_idle(host);
         err = mmc_sdio_init_card(host, host->ocr, host->card,
                     mmc_card_keep_power(host));
         ---------
}
        以下是dump_stack()打印信息：
     <4>[   55.269378] [<c0013d4c>] (unwind_backtrace+0x0/0x11c) from 
[<c057d650>] (mmc_resume_host+0xec/0x15c)
<4>[   55.269439] [<c057d650>] (mmc_resume_host+0xec/0x15c) from 
[<c058bfdc>] (msmsdcc_runtime_resume+0xb8/0x17c)
<4>[   55.269500] [<c058bfdc>] (msmsdcc_runtime_resume+0xb8/0x17c) from 
[<c058c2ac>] (msmsdcc_pm_resume+0x44/0xa8)
<4>[   55.269561] [<c058c2ac>] (msmsdcc_pm_resume+0x44/0xa8) from 
[<c03a4074>] (platform_pm_resume+0x40/0x54)
<4>[   55.269653] [<c03a4074>] (platform_pm_resume+0x40/0x54) from 
[<c03a8160>] (dpm_run_callback+0x44/0x7c)
<4>[   55.269683] [<c03a8160>] (dpm_run_callback+0x44/0x7c) from 
[<c03a8c18>] (device_resume+0x140/0x184)
<4>[   55.269744] [<c03a8c18>] (device_resume+0x140/0x184) from 
[<c03a94ec>] (dpm_resume+0xfc/0x234)
<4>[   55.269805] [<c03a94ec>] (dpm_resume+0xfc/0x234) from [<c03a97f0>] 
(dpm_resume_end+0xc/0x18)
<4>[   55.269866] [<c03a97f0>] (dpm_resume_end+0xc/0x18) from 
[<c00b63bc>] (suspend_devices_and_enter+0x240/0x314)
<4>[   55.269927] [<c00b63bc>] (suspend_devices_and_enter+0x240/0x314) 
from [<c00b65b0>] (pm_suspend+0x120/0x200)
<4>[   55.269958] [<c00b65b0>] (pm_suspend+0x120/0x200) from 
[<c00b7508>] (suspend+0x68/0x180)
<4>[   55.270019] [<c00b7508>] (suspend+0x68/0x180) from [<c0099dd8>] 
(process_one_work+0x27c/0x484)
<4>[   55.270080] [<c0099dd8>] (process_one_work+0x27c/0x484) from 
[<c009a1f0>] (worker_thread+0x210/0x3b0)
<4>[   55.270141] [<c009a1f0>] (worker_thread+0x210/0x3b0) from 
[<c009e08c>] (kthread+0x80/0x8c)
<4>[   55.270202] [<c009e08c>] (kthread+0x80/0x8c) from [<c000eaa8>] 
(kernel_thread_exit+0x0/0x8)
<4>[   55.270233] mmc2: error -110 during resume (card was removed?)

以上2点做到了理论上kernel可以正常睡眠了，下面是正常的睡眠log: